BioMed CentralBMC Ecology

ss

Open AcceResearch articleFrequency-dependent selection by wild birds promotes polymorphism in model salamandersBenjamin M Fitzpatrick*1, Kim Shook2,3 and Reuben Izally2,3

Address: 1Ecology & Evolutionary Biology, University of Tennessee, Knoxville TN 37996, USA, 2Pre-collegiate Research Scholars Program, University of Tennessee, Knoxville TN 37996, USA and 3Farragut High School, Knoxville TN 37934, USA

Email: Benjamin M Fitzpatrick* - benfitz@utk.edu; Kim Shook - theshookfamily@charter.net; Reuben Izally - izally@netzero.net

* Corresponding author

AbstractBackground: Co-occurrence of distinct colour forms is a classic paradox in evolutionary ecologybecause both selection and drift tend to remove variation from populations. Apostatic selection,the primary hypothesis for maintenance of colour polymorphism in cryptic animals, proposes thatvisual predators focus on common forms of prey, resulting in higher survival of rare forms.Empirical tests of this frequency-dependent foraging hypothesis are rare, and the link betweenpredator behaviour and maintenance of variation in prey has been difficult to confirm. Here, weshow that predatory birds can act as agents of frequency-dependent selection on terrestrialsalamanders. Polymorphism for presence/absence of a dorsal stripe is widespread in manysalamander species and its maintenance is a long-standing mystery.

Results: We used realistic food-bearing model salamanders to test whether selection by wild birdsmaintains a stripe/no-stripe polymorphism. In experimental manipulations, whichever form wasmost common was most likely to be attacked by ground-foraging birds, resulting in a survivaladvantage for the rare form.

Conclusion: This experiment demonstrates that frequency-dependent foraging by wild birds canmaintain colour polymorphism in cryptic prey.

BackgroundMaintenance of variation is a classic paradox in evolutionbecause both selection and drift tend to remove variationfrom populations [1-3]. If one form has an advantage(e.g., being more cryptic), it should replace all others.Likewise, random drift alone will eventually result in lossof all but one form when there are no fitness differences.Maintenance of a stable polymorphism requires eitherrecurrent mutation, a balance between dispersal anddivergent selection between populations, or some form ofbalancing selection within populations [3]. For example,predators are expected to form a search image for the most

common form if it is easier to search for one cryptic preytype than to simultaneously look for two [4-6]. If preda-tors switch their search images to whichever prey is mostcommon, the result is frequency-dependent selectionagainst common prey [4-7]. Empirical tests of this fre-quency-dependent selection hypothesis are rare, and thelink between predator behaviour and maintenance of var-iation in prey has been difficult to confirm [4,8]. Here, wetest whether predatory birds can act as agents of fre-quency-dependent selection on terrestrial salamandersusing a manipulative field experiment.

Published: 8 May 2009

BMC Ecology 2009, 9:12 doi:10.1186/1472-6785-9-12

Received: 10 February 2009Accepted: 8 May 2009

This article is available from: http://www.biomedcentral.com/1472-6785/9/12

© 2009 Fitzpatrick et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Page 1 of 6(page number not for citation purposes)

BMC Ecology 2009, 9:12 http://www.biomedcentral.com/1472-6785/9/12

Multiple species of small, forest floor-dwelling salaman-ders exhibit polymorphism for the presence of a dorsalstripe (Fig. 1) and what maintains this polymorphismwithin populations is a long-standing question [9-15].Polymorphism is widespread both geographically andphylogenetically, occurring in many species of Plethodonthroughout North America in addition to the distantlyrelated Batrachoseps of the Pacific coast of North America[16] and Karsenia of Asia [17]. Thus, it is unlikely that thispolymorphism is merely a transitional stage in the evolu-tion of colour pattern. Presence or absence of a dorsalstripe appears to have a simple genetic basis in Plethodoncinereus and is not related to sex [12,18].

What polymorphic salamander species seem to have incommon is that they are small, slender, and locally abun-dant. These species often comprise a substantial fractionof the animal biomass in temperate forests and are subjectto predation by ground-foraging birds and other preda-tors that search the leaf-litter for small animals [16]. Stud-ies of polymorphic Plethodon cinereus have showndifferences between the morphs in behaviour[11,14,15,19], physiology [11,13], and geographic varia-tion in relative abundance [10,20]. However, nonedirectly address how these differences might relate to amechanism maintaining variation and there is no evi-dence of similar phenotypic correlations in the other pol-ymorphic species. In fact, in P. ventralis, the frequency ofthe striped form is less at higher elevations [21] whereasthe frequency of striped P. cinereus is greater at higher lat-itude and cooler microclimates in general [10,20]. Thus,phenotypic correlations between appearance and physiol-ogy are not consistent (also see [13]) and this is not sur-prising given that dorsal colour pattern is unlikely to havea direct function in temperature tolerance or metabolism(Plethodon always avoid sunlight).

To explain the maintenance of a similar visual polymor-phism across many taxa, we propose that selection actsdirectly on appearance. The most likely function of dorsalcolour pattern is crypsis (Fig. 1), though this is anuntested assumption. Though there has been speculationthat the red stripe of P. cinereus might be an aposematicsignal [19], Brodie and Brodie [22] showed that wild birdsprey extensively upon striped P. cinereus, and give no signthat the salamanders are unpalatable. Striped P. cinereuswere taken just as often as D. ochrophaeus, which wereassumed to be cryptic and palatable [22]. A third, com-pletely red form of P. cinereus does appear to be a colourmimic of the toxic Red-spotted Newt eft (Notophthalmusviridescens), but our focus is on the more widespreadstripe/no-stripe polymorphism. Polymorphism in cryptic

Examples of polymorphic plethodontid salamandersFigure 1Examples of polymorphic plethodontid salamanders. Each pair of individuals was captured in the same place at the same time. a, Southern Red-backed Salamanders, Plethodon serratus, Great Smoky Mountains National Park (Tennessee, USA). b, Southern Zigzag Salamanders, P. ventralis, Knoxville (Tennessee, USA). c, California Slender Salamanders, Batra-choseps attenuatus, Napa County (California, USA).

Page 2 of 6(page number not for citation purposes)

BMC Ecology 2009, 9:12 http://www.biomedcentral.com/1472-6785/9/12

prey populations might be maintained in migration-selec-tion balance if alternative colour forms are adapted to dif-ferent habitats [23], or might be promoted by frequency-dependent predation causing rare form advantage withinpopulations [4,8].

Empirical tests of frequency-dependent foraging on cryp-tic prey are rare; the best-studied examples involve eithercolored pellets of dough fed to free-ranging birds[7,24,25]or digital images of moths selected by trainedBlue Jays (Cyanocitta cristata) [4,26-28]. These studiesdemonstrate that birds can generate frequency-dependentselection, that individual Blue Jays learn to focus on abun-dant prey, and that they readily switch focus in responseto changes in prey abundance. Apostatic selection by free-ranging wild birds has been shown for small pastries dif-fering in colour [7,25,29], and presence or absence of astripe [30,31], and also for real Cepaea snail shells differ-ing in banding pattern [32]. However, in other studies,selection changed from negative frequency-dependence(rare form advantage) to positive frequency-dependence(common form advantage) when artificial prey were pro-vided at high densities, showing that rare form advantageis not entirely general and may depend on abundance oraggregation of prey [25,33]. A recent study of Trinidadianguppies implicated frequency-dependent survival in themaintenance of variation in conspicuously colored males[34]. However, it is not clear whether search images areimportant for predators of conspicuous prey, and otherfactors might affect polymorphism in sexually-selectedtraits [35,36]. Thus, whether frequency-dependent forag-ing can explain the maintenance of polymorphism in real-istic prey at realistic densities remains a critical questionthat might need to be addressed on a case-by-case basis.Here, we show empirically that predatory birds can act asagents of frequency-dependent selection on terrestrial sal-amanders.

To test for frequency-dependent selection by birds, weused standardized models resembling polymorphicPlethodon salamanders (Fig. 2). A food reward (1/2 pea-nut) was glued to the underside of each model, and mod-els were set out at random in 10 × 10 m plot at the edge ofa woodlot in Knox county, Tennessee (see Methods). Wemanipulated the relative abundance of striped andunstriped prey and quantified survival by counting thenumber of models with peanuts still attached at the endof each day.

Results and DiscussionWe observed at least five individual Blue Jays (Cyanocittacristata) foraging on model salamanders. Other birdsobserved foraging on the ground in the study site includedNorthern Cardinals (Cardinalis cardinalis), American

Crows (Corvus brachyrhynchos), American Robins (Turdusmigratorius), and Eastern Towhees (Pipilio erythrophthal-mus). Only Blue Jays were observed taking peanuts fromsalamanders.

On the first test day, after unstriped models had beenmore abundant (9:1 for six days), more unstriped thanstriped model salamanders were attacked (day 7: 16/25vs. 6/25, Fisher's exact P = 0.0096). After six days with 5unstriped and 45 striped models, fewer unstriped modelswere attacked when tested at equal abundance (day 14:11/25 vs. 19/25, Fisher's exact P = 0.0421). Whicheverform had been rare (1:9) over the past six days had a sig-nificant survival advantage over the common form whenpresented at equal abundance (Table 1).

Experimental subjectsFigure 2Experimental subjects. (a) Polymorphic model salaman-ders. (b) Blue Jay (Cyanocitta cristata) foraging in the study area.

Page 3 of 6(page number not for citation purposes)

BMC Ecology 2009, 9:12 http://www.biomedcentral.com/1472-6785/9/12

Fig. 3 shows that striped models had higher relative fitnesswhen at low frequency and lower relative fitness when athigh frequency. Multiple logistic regression supported asignificant relationship between frequency and survivalon non-test days (likelihood ratio G2 = 6.37, df = 1, P =0.0116). There was also a strong effect of "day" (G2 =58.27, df = 12, P < 0.0001), which probably reflects varia-tion in predator activity (predation was lighter on rainydays) and a delay in learning. There was no detectableassociation between survival and morph (likelihood ratioG2 = 0.225, df = 1, P = 0.635). Whichever form was rarehad a survival advantage within one or two days after therelative abundances were reversed (Fig. 3). Thus, whenwild birds were "trained" to expect one form to be mostcommon, they preyed disproportionately on that form.This occurred regardless of which form was common.

Selection by birds can be heterogeneous and these resultsshould be tested at other sites. However, decades ofresearch have consistently found that captive and free-ranging wild birds cause frequency-dependent selectionon all manner of food items differing in appearance[4,6,7,25,26,28]. We show that this very general processcan operate on the kind of phenotypic polymorphismseen commonly in small salamanders.

We have not tested the perceptual and behavioural mech-anisms underlying frequency-dependent foraging, there-fore we have avoided invoking search image formation asthe specific ultimate cause of rare form advantage. Never-theless, this study joins a growing body of literature show-ing that birds often switch foraging behaviour in responseto the relative abundance of alternative cryptic prey andthis can promote the maintenance of colour polymor-phism [4,8]. The cognitive processes underlying this phe-nomenon are important subjects for further research.

In addition, a crucial question for the maintenance of pol-ymorphism is to what extent frequency-dependent forag-ing by visual predators is counteracted by other selectiveprocesses, including non-visual predation. Other sourcesof mortality and other phenotypic differences betweenmorphs might interfere with or enhance the stable coexist-ence of distinct colour morphs. The conceptual linkbetween promoting polymorphism and promoting coex-istence of similar species [37] merits greater attention.

ConclusionRare form advantage owing to frequency-dependent selec-tion by visual predators has been proposed to explainmaintenance of polymorphism in many cryptic species[4]. The experiment reported here provides evidence thatthe key mechanism (frequency-dependent foraging) oper-ates with realistic prey at realistic densities in the wild. Themaintenance of colour variation in terrestrial salamandersmight be explained by the oldest and most obvioushypothesis – rare form advantage arises because predatorstend to overlook rare prey.

MethodsModel salamanders were constructed with modelling clay(Sculpey™). One model was sculpted by hand and used tocreate a flexible mould. This mould was then used tomake models of consistent size and body position. Mod-els were either grey-brown or grey-brown with an ochrestripe from the base of the head to the base of the tail (Fig.2a). A food reward was associated with each model sala-mander by adhering one peanut half to the bottom of themodel using edible library paste (40 g corn starch, 1 Lwater, boiled and cooled).

Table 1: "Survival" rates of model salamanders. Data are given as number survived/number presented each day.

Datea Stripe No Stripe

6 Jul 1/5 17/457 Jul 4/5 21/458 Jul 4/5 20/459 Jul 4/5 30/4510 Jul 5/5 35/4511 Jul 4/5 24/4512 Jul 19/25 9/2513 Jul 35/45 3/514 Jul 25/45 3/515 Jul 16/45 3/516 Jul 26/45 4/517 Jul 26/45 4/518 Jul 25/45 4/519 Jul 6/25 14/25

a 2008 AD

Relative fitness of model salamandersFigure 3Relative fitness of model salamanders. Daily survival rates of striped relative to unstriped models (right-hand axis, filled symbols) in comparison to the frequency of striped models (left-hand axis, open symbols). Dashed line illustrates equal frequency and equal fitness.

Page 4 of 6(page number not for citation purposes)

BMC Ecology 2009, 9:12 http://www.biomedcentral.com/1472-6785/9/12

The field site was an open area adjacent to a woodlot inKnox County, TN. Polymorphic P. ventralis are knownfrom within 2 km. To avoid spurious loss of models, weused as foraging units 100 black plastic trays (0.185 m2)placed in a 10 × 10 array and filled with partially decom-posed leaf-litter (natural forest floor substrate). One m ofopen lawn was left between adjacent trays. Each day, weused a random number generator to assign model sala-manders to 50 trays. The ordering of striped and unstripedmodels was also random; as a result, each tray positionhad a 50% chance of having a model salamander each dayand the probability that the model was striped was the fre-quency of striped models for that day. The resulting den-sity of 1/2 m-2 was realistic for small Plethodon andBatrachoseps salamanders [16]. Models were placed hap-hazardly within the trays and arranged so that theattached peanut could not be seen without moving themodel. This design ensured that foraging birds had theopportunity to learn where to look for salamanders andthat all "surviving" salamanders would be recovered at theend of each day. In addition, it was important to makesure that each tray was not guaranteed to provide a sala-mander, so that sometimes giving up and moving toanother was the right foraging decision [27].

The experiment was divided into two training periods ofsix days each and two test days, similar to protocols usedwith green and brown pastry pellets[7,25]. First, 5 stripedand 45 unstriped salamander models were presented eachday from 8:00 am until 7:30 pm (it usually took this entiretime for the birds to attack roughly half of the salamandermodels). At the end of the day, the flats were gathered andthe number of "surviving" salamanders counted (thosewith the peanut still attached). Six days of training werefollowed immediately by a test day in which 25 of eachmorph were presented. After the first test day, another six-day training period was started, this time with frequenciesof 45 striped and 5 unstriped salamander models. The sec-ond test day (25 of each morph) immediately followedthe second training period (Table 1).

We used 2 × 2 contingency tables to test the null hypoth-esis of equal predation rates for striped and unstripedmodels on the two test days. For the 12 "training" days, wetested for an effect of relative abundance on survival usinglogistic regression with individual survival as the responsevariable, day as a random effect, and morph and fre-quency category as fixed effects. The goodness-of-fit of themodel was adequate (likelihood ratio test relative to a sat-urated model: G2 = 4.03, df = 11, P = 0.708) and addinginteraction terms did not significantly improve the fit. Forall 14 days of the experiment, we estimated fitness (sur-vival value) of striped relative to unstriped models as wS =[(SA + 0.5)/(SB + 0.5)] ÷ [(UA + 0.5)/(UB + 0.5)], where SAand UA are the numbers of surviving striped and unstriped

models, and SB and UB are the numbers before selection.This estimates the ratio of the survival probability ofstriped relative to unstriped models [38]. For example,when the two forms have equal fitness, wS = 1.0 and whena striped individual is twice as likely to survive as anunstriped individual, wS = 2.0.

Authors' contributionsBMF Conceived the project, collected and analyzed thedata, and drafted the manuscript. KS and RI helped trou-bleshoot the experimental design, collect and analyzedata, and revise the manuscript. All authors read andapproved the final manuscript.

AcknowledgementsWe thank R. Fitzpatrick for help with models and fieldwork; and J. Allen, N. Sanders, J. Fordyce, and one anonymous reviewer for comments on the manuscript. This project was supported by the University of Tennessee Pre-collegiate Research Scholars Program.

References1. Ford EB: Ecological genetics. London: Methuen and Co.; 1964. 2. Futuyma DJ: Evolution. Sunderland, MA: Sinauer Associates, Inc.;

2005. 3. Hartl DL, Clark AG: Principles of population genetics. 3rd edi-

tion. Sunderland, MA: Sinauer Associates, Inc.; 1997. 4. Bond AB: The evolution of color polymorphism: Crypticity,

searching images, and apostatic selection. Annual Review ofEcology, Evolution and Systematics 2007, 38:489-514.

5. Poulton EB: The colours of animals: Their meaning and use,especially considered in the case of insects. New York: Apple-ton; 1890.

6. Tinbergen L: The natural control of insects on pinewoods. I.Factors influencing the intensity of predation by songbirds.Arch Neerlandaises Zool 1960, 13:265-343.

7. Allen JA, Clarke BC: Evidence for apostatic selection by wildpasserines. Nature 1968, 220:501-502.

8. Punzalan D, Rodd FH, Hughes KA: Perceptual processes and themaintenance of polymorphism through frequency-depend-ent predation. Evolutionary Ecology 2005, 19:303-320.

9. Burger JW: Plethodon cinereus (Green) in eastern Pennsylva-nia and New Jersey. American Naturalist 1935, 64:578-586.

10. Lotter F, Scott NJ: Correlation between climate and distribu-tion of the color morphs of the salamander Plethodon cin-ereus. Copeia 1977, 1977:681-690.

11. Moreno G: Behavioral and physiological differentiationbetween the color morphs of the salamander, Plethodon cin-ereus. Journal of Herpetology 1989, 3:335-341.

12. Highton R: Geographic variation in genetic dominance ofcolor morphs of Red-backed Salamander, Plethodon cinereus.Genetics 1975, 80:363-374.

13. Petruzzi EE, Niewiarowski PH, Moore FB-G: The role of thermalniche selection in maintenance of a colour polymorphism inredback salamanders (Plethodon cinereus). Frontiers in Zoology2006, 3:10.

14. Anthony CD, Venesky MD, Hickerson C-AM: Ecological separa-tion in a polymorphic terrestrial salamander. Journal of AnimalEcology 2008, 77:646-653.

15. Test FH: Seasonal differences in populations of the red-backed salamander in southeastern Michigan. Papers of theMichigan Academy of Science, Arts, and Letters 1955, 40:137-153.

16. Petranka JW: Salamanders of the United States and Canada.Washington, D. C.: Smithsonian Institution Press; 1998.

17. Min MS, Yang SY, Bonnett RM, Vieites DR, Brandon RA, Wake DB:Discovery of the first Asian plethodontid salamander. Nature2005, 435:87-90.

18. Highton R: The inheritance of the color phases of Plethodoncinereus. Copeia 1959, 1:33-37.

Page 5 of 6(page number not for citation purposes)

BMC Ecology 2009, 9:12 http://www.biomedcentral.com/1472-6785/9/12

Publish with BioMed Central and every scientist can read your work free of charge

"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."

Sir Paul Nurse, Cancer Research UK

Your research papers will be:

available free of charge to the entire biomedical community

peer reviewed and published immediately upon acceptance

cited in PubMed and archived on PubMed Central

yours — you keep the copyright

Submit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.asp

BioMedcentral

19. Venesky MD, Anthony CD: Antipredator adaptations and pred-ator avoidance by two color morphs of the Eastern Red-backed Salamander, Plethodon cinereus. Herpetologica 2007,63:450-458.

20. Gibbs JP, Karraker NE: Effects of warming conditions in EasternNorth American forests on Red-backed Salamander mor-phology. Conservation Biology 2006, 20:913-917.

21. Highton R: Geographic protein variation and speciation in thePlethodon dorsalis complex. Herpetologica 1997, 53:345-356.

22. Brodie ED Jr, Brodie ED III: Differential avoidance of mimeticsalamanders by free-ranging birds. Science 1980, 208:181-182.

23. Nosil P: Reproductive isolation caused by visual predation onmigrants between divergent environments. Proceedings of theRoyal Society of London, Series B 2004, 271:1521-1528.

24. Allen JA: Frequency-dependent selection by predators. Philo-sophical Transactions of the Royal Society of London Series B 1988,319:485-503.

25. Weale ME, Whitwell D, Raison HE, Raymond DL, Allen JA: Theinfluence of density on frequency-dependent food selection:A comparison of four experiments with wild birds. Oecologia2000, 124:391-395.

26. Bond AB, Kamil AC: Apostatic selection by blue jays producesbalanced polymorphism in virtual prey. Nature 1998,395:594-596.

27. Bond AB, Kamil AC: Visual predators select for crypticity andpolymorphism in virtual prey. Nature 2002, 415:609-614.

28. Bond AB, Kamil AC: Spatial heterogeneity, predator cognition,and the evolution of color polymorphism in virtual prey. Pro-ceedings of the National Academy of Sciences, USA 2006, 103:3214-3219.

29. Allen JA: Wild birds prefer to eat the more familiar of artificialmorphs that are similar in color. Heredity 1984, 53:705-715.

30. Raymond DL: Wild birds prefer the familiar of striped andunstriped artificial prey. Biological Journal of the Linnean Society1984, 23:229-235.

31. Raymond DL, Allen JA: Wild birds prefer the familiar color pat-tern when feeding on similar artificial morphs. Oikos 1990,57:175-179.

32. Allen JA, Raymond DL, Geburtig MA: Wild birds prefer the famil-iar morph when feeding on pastry-filled shells of the landsnailCepaea hortensis (Mull). Biological Journal of the Linnean Society1988, 33:395-401.

33. Allen JA, Weale ME: Anti-apostatic selection by wild birds onquasi-natural morphs of the land snail Cepaea hortensis : ageneralised linear mixed models approach. Oikos 2005,108:335-343.

34. Olendorf R, Rodd FH, Punzalan D, Houde AE, Hunt C, Reznick DN,Hughes KA: Frequency-dependent survival in natural guppypopulations. Nature 2006, 441:633-636.

35. Sinervo B, Calsbeek R: The developmental, physiological, neu-ral, and genetical causes and consequences of frequency-dependent selection in the wild. Annual Review of Ecology, Evolu-tion and Systematics 2006, 37:581-610.

36. Sinervo B, Lively C: The rock-paper-scissors game and the evo-lution of alternative male strategies. Nature 1996, 380:240-243.

37. Clarke BC: Balanced polymorphism and the diversity of sym-patric species. In Taxonomy and geography Edited by: Nichols D.Oxford: Systematics Assoc; 1962:47-70.

38. Haldane JBS: The estimation and significance of the logarithmof the ratio of frequencies. Annals of Human Genetics 1956,20:309-311.

Page 6 of 6(page number not for citation purposes)